System, method, and computer program product for iteratively refining a training data set

A system, comprising:

• at least one processor programmed or configured to:
• train a first predictive model based on a first set of user profiles, the first set of user profiles comprising:
• a first subset of user profiles associated with a first profile type, and
• a second subset of user profiles associated with a second profile type;
• determine a classification for each user profile of a second set of user profiles using the first predictive model, the classification indicating whether each user profile is associated with the first profile type or the second profile type;
• determine a performance score for the first predictive model, wherein the performance score indicates an accuracy of the first predictive model when classifying each user profile of the second set of user profiles;
• determine whether to update the first predictive model based on the performance score for the first predictive model;
• determine a classification for each user profile of the first set of user profiles using the first predictive model;
• select at least one user profile of the first subset of user profiles or at least one user profile of the second subset of user profiles to include in a removal set of user profiles based on the classification for each user profile of the first set of user profiles;
• remove each user profile included in the removal set of user profiles from the first set of user profiles;
• receive data associated with a third set of user profiles;
• determine whether each user profile of the third set of user profiles is associated with the first profile type or the second profile type using the first predictive model;
• and
• transmit data associated with each user profile of the third set of user profiles.

A computer program product comprising at least one non-transitory computer-readable medium including one or more instructions that, when executed by at least one processor, cause the at least one processor to:

• train a first predictive model based on a first set of user profiles, the first set of user profiles comprising:
• a first subset of user profiles associated with a first profile type, and
• a second subset of user profiles associated with a second profile type;
• determine a classification for each user profile of a second set of user profiles using the first predictive model, the classification indicating whether each user profile is associated with the first profile type or the second profile type;
• determine a performance score for the first predictive model, wherein the performance score indicates an accuracy of the first predictive model when classifying each user profile of the second set of user profiles;
• determine whether to update the first predictive model based on the performance score for the first predictive model;
• determine a classification for each user profile of the first set of user profiles using the first predictive model;
• select at least one user profile of the first subset of user profiles or at least one user profile of the second subset of user profiles to include in a removal set of user profiles based on the classification for each user profile of the first set of user profiles;
• remove each user profile included in the removal set of user profiles from the first set of user profiles;
• receive data associated with a third set of user profiles;
• determine whether each user profile of the third set of user profiles is associated with the first profile type or the second profile type using the first predictive model;
• and
• transmit data associated with each user profile of the third set of user profiles.